Mostrar el registro sencillo del ítem

dc.contributor.authorGómez Camperos, July
dc.contributor.authorRuiz Diaz, Carlos Mauricio
dc.contributor.authorHernanez Cely, Marlon Mauricio
dc.contributor.authorHernandez Rodriguez, Oscar Mauricio
dc.contributor.authorPardo Garcia, Aldo
dc.coverage.spatialColombiaen_US
dc.date.accessioned2026-04-22T16:39:42Z
dc.date.available2026-04-22T16:39:42Z
dc.date.issued2025-10-08
dc.identifier.issn1806-3691en_US
dc.identifier.urihttps://repositorioinstitucional.ufpso.edu.co/xmlui/handle/20.500.14167/6224
dc.description.abstractLa identificación precisa de los patrones de flujo bifásico de gas denso y líquido es esencial en los sistemas de transporte de fluidos por tuberías y en las operaciones de los pozos petrolíferos. Sin embargo, los métodos convencionales para clasificar estos patrones se basan principalmente en la observación visual directa y en sensores instalados en el sistema. Este estudio presenta un enfoque innovador para el reconocimiento de patrones de flujo, basado en algoritmos de redes neuronales convolucionales (CNN) modificadas y en el aprendizaje por transferencia basado en técnicas de regularización, con el fin de lograr una identificación automática y objetiva de los patrones en flujos de gas denso/líquido. Los experimentos se llevaron a cabo en la Plataforma Experimental de Flujo Multifásico del Laboratorio de Flujo Multifásico Industrial (LEMI). Se identificaron seis patrones de flujo en tuberías horizontales, lo que generó una base de datos de 67 710 imágenes, que se dividen en tres categorías: el 60 % se utiliza para el entrenamiento del modelo, el 20 % para la validación y el 20 % restante para la evaluación del modelo. El modelo propuesto se evaluó en dieciocho arquitecturas de redes neuronales convolucionales modificadas y en seis conjuntos de prueba. Los resultados computacionales mostraron que el modelo propuesto fue capaz de identificar correctamente los patrones de flujo bifásico con una precisión de 1,0 con las arquitecturas MobileNet, InceptionV3 e InceptionResNetV2, en el cuarto conjunto de prueba, destacando en rendimiento entre los seis conjuntos evaluados sin presentar señales de sobreajuste. Por el contrario, los modelos ResNet50, ResNet152V2, MobileNetV3Small, MobileNetV3Large, ResNet101V2, InceptionV4, AlexNet y Linet5 no lograron alcanzar una precisión superior a 0,2, obteniendo los resultados más bajos en los conjuntos de entrenamiento.en_US
dc.language.isoengen_US
dc.relationhttps://www.sciencedirect.com/science/article/pii/S0301932225003544en_US
dc.titleReconocimiento de patrones de flujo bifásico gas/líquido denso mediante redes neuronales convolucionales y aprendizaje por transferencia.en_US
dc.title.translatedDense-gas/liquid two-phase flow pattern recognition using convolutional neural networks and transfer learningen_US
dc.contributor.cvlacJuly Gómez-Camperos [0001164287]en_US
dc.contributor.cvlacCarlos Mauricio Ruiz Diaz [0001772117]en_US
dc.contributor.cvlacMarlon Mauricio Hernanez Cely [0001354455]en_US
dc.contributor.cvlacOscar Mauricio Hernandez Rodriguez [0000316504]en_US
dc.contributor.cvlacAldo Pardo Garcia [0000173878]en_US
dc.contributor.googlescholarCarlos Mauricio Ruiz Diaz [x10qI0UAAAAJ]en_US
dc.contributor.googlescholarMarlon Mauricio Hernanez Cely [RQyKQW4AAAAJ]en_US
dc.contributor.googlescholarOscar Mauricio Hernandez Rodriguez [i48N_sEAAAA]en_US
dc.contributor.orcidJuly Gómez Camperos [0000-0003-3894-144X]en_US
dc.contributor.orcidCarlos Mauricio Ruiz Diaz [0000-0003-3836-9332]en_US
dc.contributor.orcidMarlon Mauricio Hernanez Cely [0000-0001-7564-4044]en_US
dc.contributor.orcidOscar Mauricio Hernandez Rodriguez [0000-0003-1578-7996]en_US
dc.contributor.researchgateCarlos Mauricio Ruiz Diaz [Carlos-Mauricio-Ruiz-Diaz]en_US
dc.contributor.scopusJuly Gómez Camperos [57220550902]en_US
dc.contributor.scopusCarlos Mauricio Ruiz Diaz [57326268000]en_US
dc.contributor.scopusMarlon Mauricio Hernanez Cely [57195293394]en_US
dc.contributor.scopusOscar Mauricio Hernandez Rodriguez [7103218937]en_US
dc.contributor.scopusAldo Pardo Garcia [57884427900]en_US
dc.description.abstractenglishAccurate identification of dense-gas/liquid two-phase flow patterns is essential in pipeline fluid transport systems and oil well operations. However, conventional methods for classifying these patterns rely mostly on direct visual observation and sensors installed in the system. This study presents an innovative approach for flow pattern recognition, based on modified convolutional neural network (CNN) algorithms and transfer learning based on regularization techniques to achieve an automatic and objective identification of patterns in dense-gas/liquid flows. Experiments were carried out in the Multiphase Flow Experimental Platform of the Industrial Multiphase Flow Laboratory (LEMI). Six flow patterns in horizontal pipes were identified, generating a database of 67,710 images, which are divided into three categories: 60 % are used for model training, 20 % for validation and the remaining 20 % for model evaluation. The proposed model was evaluated on eighteen modified convolutional neural network architectures, and in six test sets. The computational results showed that the proposed model was able to identify the two-phase flow patterns correctly with an accuracy of 1.0 with MobileNet, InceptionV3, InceptionResNetV2 architectures, in the fourth test set, standing out in performance among the six evaluated sets without presenting over-fitting signals. In contrast, the ResNet50, ResNet152V2, MobileNetV3Small, MobileNetV3Large, ResNet101V2, InceptionV4, AlexNet and Linet5 models failed to achieve an accuracy higher than 0.2, obtaining the lowest results in the training sets.en_US
dc.subject.proposalFlujo de gas denso/líquidoen_US
dc.subject.proposalFlujo bifásicoen_US
dc.subject.proposalRed neuronal convolucional (CNN)en_US
dc.subject.proposalClasificación de imágenesen_US
dc.subject.proposalAprendizaje por transferenciaen_US
dc.subject.keywordsDense-gas/liquid flowen_US
dc.subject.keywordsTwo-phase flowen_US
dc.subject.keywordsConvolutional neural network (CNN)en_US
dc.subject.keywordsImage classificationen_US
dc.subject.keywordsTransfer learningen_US
dc.publisher.facultyFacultad ingenieríasen_US
dc.type.redcolhttp://purl.org/redcol/resource_type/ARTen_US
dc.relation.referencesAL-Dogail, A.S., Gajbhiye, R.N., 2021. Effects of density, viscosity and surface tension on flow regimes and pressure drop of two-phase flow in horizontal pipes. J. Pet. Sci. Eng. 205 (March), 108719. https://doi.org/10.1016/j.petrol.2021.108719en_US
dc.relation.referencesAL-Dogail, A.S., Gajbhiye, R.N., 2021. Effects of density, viscosity and surface tension on flow regimes and pressure drop of two-phase flow in horizontal pipes. J. Pet. Sci. Eng. 205 (April), 108719. https://doi.org/10.1016/j.petrol.2021.108719.en_US
dc.relation.referencesArabi, A., Salhi, Y., Zenati, Y., Si-Ahmed, E.K., Legrand, J., 2020. On gas-liquid intermittent flow in a horizontal pipe: influence of sub-regime on slug frequency. Chem. Eng. Sci. 211, 115251. https://doi.org/10.1016/j.ces.2019.115251.en_US
dc.relation.referencesBelt, R., Smith, I.E., Staff, G., Lawrence, C., 2024. Growth of solitary slugs in long pipes. Int. J. Multiph. Flow 178 (May), 104883. https://doi.org/10.1016/j. ijmultiphaseflow.2024.104883.en_US
dc.relation.referencesCarneiro, J.N.E., et al., Oct. 2015. Numerical simulations of high CO2 content flows in production wells, flowlines and risers. In: Proceedings of the OTC Brasil, OTC, pp. 1522–1533. https://doi.org/10.4043/26231-MS.en_US
dc.relation.referencesdas Neves, D.A., Vieira, S.C., Cenzi, J.R., de Paula, A.S., de Castro, M.S., Fabro, A.T., 2024. Identification of the flow pattern from the experimental pressure signal in horizontal pipes carrying two-phase flows. Exp. Therm. Fluid Sci. 154 (March 2023), 111141. https://doi.org/10.1016/j.expthermflusci.2024.111141.en_US
dc.relation.referencesDong, F., Zhang, S., Shi, X., Wu, H., Tan, C., 2021. Flow regimes identification-based multidomain features for gas-liquid two-phase flow in horizontal pipe. IEEE Trans. Instrum. Meas. 70, 1–11. https://doi.org/10.1109/TIM.2021.3089767.en_US
dc.relation.referencesGonçalves, J.L., Mazza, R.A., 2024. Slug flow parameters applied to dynamic riser analysis. Int. J. Multiph. Flow 171 (November 2023). https://doi.org/10.1016/j. ijmultiphaseflow.2023.104670en_US
dc.relation.referencesHe, K., Zhang, X., Ren, S., Sun, J., 2016. Identity mappings in deep residual networks. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics) 9908, 630–645. https://doi.org/10.1007/978-3-319-46493-0_38. LNCS.en_US
dc.relation.referencesHoward A.G., et al., “MobileNets: efficient convolutional neural networks for mobile vision applications,” 2017.en_US
dc.relation.referencesHuang, G., Liu, Z., Van Der Maaten, L., Weinberger, K.Q., 2017. Densely connected convolutional networks. In: Proceedings of the 30th IEEE Conference on Computer Vision Pattern Recognition, CVPR 2017, 2017-Janua, pp. 2261–2269. https://doi. org/10.1109/CVPR.2017.243.en_US
dc.relation.referencesKhan, U., Pao, W., Pilario, K.E., Sallih, N., 2024. Flow regime classification using various dimensionality reduction methods and AutoML. Eng. Anal. Bound. Elem. 163 (November 2023), 161–174. https://doi.org/10.1016/j.enganabound.2024.03.006.en_US
dc.relation.referencesKuang, B., Nnabuife, S.G., Whidborne, J.F., Sun, S., Zhao, J., Jenkins, K., 2024. Selfsupervised learning-based two-phase flow regime identification using ultrasonic sensors in an S-shape riser. Expert. Syst. Appl. 236 (August 2023), 121414. https:// doi.org/10.1016/j.eswa.2023.121414.en_US
dc.relation.referencesLiu, W., Tan, C., Dong, F., 2021. Doppler spectrum analysis and flow pattern identification of oil-water two-phase flow using dual-modality sensor. Flow Meas. Instrum. 77 (March 2020), 101861. https://doi.org/10.1016/j. flowmeasinst.2020.101en_US
dc.relation.referencesLuo, C., et al., 2023. Experimental and modeling investigation on gas-liquid two-phase flow in horizontal gas wells. J. Energy Resour. Technol. Trans. ASME 145 (1), 1–13. https://doi.org/10.1115/1.4055223.en_US
dc.relation.referencesMiseta, T., Fodor, A., Vathy-Fogarassy, A., ´ 2024. Surpassing early stopping: a novel correlation-based stopping criterion for neural networks. Neurocomputing 567 (November 2023), 127028. https://doi.org/10.1016/j.neucom.2023.127028.en_US
dc.relation.referencesMohammed, S., Abdulkareem, L., Roshani, G.H., Eftekhari-Zadeh, E., Haso, E., 2022. Enhanced multiphase flow measurement using dual non-intrusive techniques and ANN model for void fraction determination. Processes 10 (11). https://doi.org/ 10.3390/pr10112en_US
dc.relation.referencesMohmmed, A.O., Al-Kayiem, H.H., Osman, A.B., 2021. Investigations on the slug twophase flow in horizontal pipes: past, presents, and future directives. Chem. Eng. Sci. 238, 116611. https://doi.org/10.1016/j.ces.2021.116611.en_US
dc.relation.referencesMumuni.A, and F. Mumuni, “Data augmentation with automated machine learning: approaches and performance comparison with classical data augmentation methods,” arXiv.org, abs/2403.08352, pp. 1–28, 2024.en_US
dc.relation.referencesNakamura, K., Derbel, B., Won, K.J., Hong, B.W., 2021. Learning-rate annealing methods for deep neural networks. Electron 10 (16), 1–12. https://doi.org/10.3390/ electronics10162en_US
dc.relation.referencesNie, F., Wang, H., Song, Q., Zhao, Y., Shen, J., Gong, M., 2022. Image identification for two-phase flow patterns based on CNN algorithms. Int. J. Multiph. Flow 152 (February), 104067. https://doi.org/10.1016/j.ijmultiphaseflow.2022.104067.en_US
dc.relation.referencesNnabuife, S.G., Kuang, B., Whidborne, J.F., Rana, Z.A., 2023. Development of gas-liquid flow regimes identification using a noninvasive ultrasonic sensor, belt-shape features, and convolutional neural network in an S-shaped riser. IEEe Trans. Cybern. 53 (1), 3–17. https://doi.org/10.1109/TCYB.2021.3084860.en_US
dc.relation.referencesOnuoha, M.D.U., Duan, M., Xu, X., Liu, J., 2022. Numerical modelling of the interaction between severe slugging and dynamic response of deepwater risers: a time-domain ANHE-FD based solution. Int. J. Non Linear Mech. 138 (July 2021), 103834. https:// doi.org/10.1016/j.ijnonlinmec.2021.103834.en_US
dc.relation.referencesQiao, W., Guo, H., Huang, E., Chen, H., Lian, C., 2023. Two-phase flow pattern identification by embedding double attention mechanisms into a convolutional neural network. J. Mar. Sci. Eng. 11 (4). https://doi.org/10.3390/jmse11040793.en_US
dc.relation.referencesQiumei, Z., Yukun, H., Fenghua, W., Zhang, P., Chao, L., 2023. Liqnet: a real-time monitoring network for two-phase flow patterns. Flow Meas. Instrum. 90 (December 2022), 102313. https://doi.org/10.1016/j.flowmeasinst.2023.102313.en_US
dc.relation.referencesQuintino, A.M., da Fonseca Junior, R., Rodriguez, O.M.H., 2023. Experimental study of liquid/dense-gas pipe flow. Geoenergy Sci. Eng. 230 (March). https://doi.org/ 10.1016/j.geoen.en_US
dc.relation.referencesSalgado, C.M., de F. Dam, R.S., de C. Conti, C., Salgado, W.L., 2021. Three-phase flow meters based on X-rays and artificial neural network to measure the flow compositions. Flow Meas. Instrum. 82 (October). https://doi.org/10.1016/j. flowmeasinst.2021.102075.en_US
dc.relation.referencesSandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C., 2017. MobileNetV2: inverted residuals and linear bottlenecks. Proc. IEEE Conf. Comput. Vis. Pattern Recognit. 4510–4520. https://doi.org/10.1109/CVPR.2018.00474.en_US
dc.relation.referencesShi, X., Tan, C., Dong, F., Santos, E.N.D., Da Silva, M.J., 2021. Conductance sensors for multiphase flow measurement: a review. IEEE Sens. J. 21 (11), 12913–12925. https://doi.org/10.1109/JSEN.2020.3042206.en_US
dc.relation.referencesShin, H.C., Kim, S.H., Shah, Y., Kim, S.M., 2024. An experimental study on air-oil flow patterns in horizontal pipes using two synthetic oils. Int. J. Heat. Mass Transf. 226 (March), 125459. https://doi.org/10.1016/j.ijheatmasstransfer.2024.125459. Simonyan, K., Zisserman, A., 2015. Very deep convolutional networks for large-scale image recognition. In: Proceedings of the 3rd International Conference on Learning Representations ICLR 2015 - Conference Track, pp. 1–14.en_US
dc.relation.referencesTandoh, H., Nnabuife, S.G., Cao, Y., Lao, L., Whidborne, J., 2022. Slug flow regime in a flowline with U-shape riser. Case Stud. Chem. Environ. Eng. 6 (August), 100265. https://doi.org/10.1016/j.cscee.2022.100265.en_US
dc.relation.referencesVaz, R.G., von Hohendorff Filho, J.C., Botechia, V.E., Schiozer, D.J., October 2021. Modeling subsea gas-liquid separation using proxy for phase separation for pre-salt reservoirs with high GOR. J. Pet. Sci. Eng. 212, 2022. https://doi.org/10.1016/j. petrol.2022.110267en_US
dc.relation.referencesWang, Y., Yu, Y., Liu, Z., Chang, Y., Zhao, X., Wang, Q., 2024. Experimental study on gasliquid two-phase stratified flow at high pressure in a horizontal pipe. Energies (Basel) 17 (5). https://doi.org/10.3390/en17051056.en_US
dc.relation.referencesXu, H., Tang, T., Zhang, B., Liu, Y., 2022. Identification of two-phase flow regime in the energy industry based on modified convolutional neural network. Prog. Nucl. Energy 147 (April 2021), 104191. https://doi.org/10.1016/j.pnucene.2022.104191.en_US
dc.relation.referencesYan, Q., Li, D., Wang, K., Zheng, G., 2024. Study on the hydrodynamic evolution mechanism and drift flow patterns of pipeline gas–liquid flow. Processes 12 (4). https://doi.org/10.3390/pr12040695.en_US
dc.relation.referencesYang, G., et al., 2024. Numerical analysis on two-phase flow-induced vibrations at different flow regimes in a spiral tube. Nucl. Eng. Technol. 56 (5), 1712–1724. https://doi.org/10.1016/j.net.2023.12.025.en_US
dc.relation.referencesYaqub, M.W., Pendyala, R., 2024. Flow regime transition maps and pressure loss prediction of gas, oil and water three-phase flow in the vertical riser downstream 90◦ bend using data driven approach. Digit. Chem. Eng. 12 (April), 100174. https://doi. org/10.1016/j.dche.2024.100174.en_US
dc.relation.referencesYuan, X., Feng, Z., Norton, M., Li, X., 2019. Generalized batch normalization: towards accelerating deep neural networks. In: Proceedings of the 33rd AAAI Conference on Artificial Intelligence AAAI 2019, 31st Innovative Applications of Artificial Intelligence Conference IAAI 2019 9th AAAI Symposium on Educational Advances in Artificial Intelligence EAAI 2019, pp. 1682–1689. https://doi.org/10.1609/aaai. v33i01.33011682.en_US
dc.relation.referencesZhang, D., Ouyang, B., Luo, Z.H., 2024. Identification of gas-solid flow regimes using convolutional neural network techniques. Powder Technol. 442 (March), 119848. https://doi.org/10.1016/j.powtec.2024.119848.en_US
dc.relation.referencesZhang, J., Wei, X., Wang, Z., 2023. The recognition algorithm of two-phase flow patterns based on GoogLeNet+5 coord attention. Micromachines (Basel) 14 (2). https://doi. org/10.3390/mi14en_US
dc.relation.referencesZhao, X., Wang, L., Zhang, Y., Han, X., Deveci, M., Parmar, M., 2024. A Review of Convolutional Neural Networks in Computer Vision, 57. Springer, Netherlands. https://doi.org/10.1007/s10462-024-10721-6.en_US
dc.relation.referencesZoph, B., Vasudevan, V., Shlens, J., Le, Q.V., 2018. Learning transferable architectures for scalable image recognition. Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern. Recognit. 8697–8710. https://doi.org/10.1109/CVPR.2018.00907en_US
dc.relation.referencesZou, S., Guo, L., Yao, T., 2021. Upstream-flow-based mechanisms for global flow regime transition of gas/liquid two-phase flow in pipeline-riser systems. Chem. Eng. Sci. 240, 116542. https://doi.org/10.1016/j.ces.2021.116542.en_US
dc.identifier.DOIhttps://doi.org/10.1007/s40430-025-06289-zen_US


Ficheros en el ítem

Thumbnail

Este ítem aparece en la(s) siguiente(s) colección(ones)

Mostrar el registro sencillo del ítem